lable at ScienceDirect

Atmospheric Environment 115 (2015) 153e162

Contents lists avai

Atmospheric Environment

journal homepage: www.elsevier .com/locate/atmosenv

Characteristics and reactivity of volatile organic compounds fromnon-coal emission sources in China

Qiusheng He a, *, Yulong Yan a, Hongyan Li a, Yiqiang Zhang b, Laiguo Chen b, **,Yuhang Wang c

a School of Environment and Safety, Taiyuan University of Science and Technology, Taiyuan, Chinab Center of Urban Air Pollution, South China Institute of Environmental Science (SCIES), Ministry of Environmental Protection (MEP), Guangzhou, Chinac School of Earth and Atmospheric Science, Georgia Institute of Technology, Atlanta, GA, USA

h i g h l i g h t s

� Abundant VOCs species profiles for non-coal sources in China were presented.� The source profiles developed in our study showed some differences with others reported.� B/T is a good marker to distinguish some VOCs sources but not all.� Most of the VOCs emissions from non-coal sources have high air reactivity.

a r t i c l e i n f o

Article history:Received 22 August 2014Received in revised form26 May 2015Accepted 28 May 2015Available online 30 May 2015

Keywords:Volatile organic compounds (VOCs)Source profilesBTEX ratiosChemical reactivity

* Corresponding author.** Corresponding author.

E-mail addresses: heqs@tyust.edu.cn (Q. He), chen

http://dx.doi.org/10.1016/j.atmosenv.2015.05.0661352-2310/© 2015 Elsevier Ltd. All rights reserved.

a b s t r a c t

Volatile organic compounds (VOCs) were sampled from non-coal emission sources including fuel refu-eling, solvent use, industrial and commercial activities in China, and 62 target species were determinedby gas chromatography-mass selective detector (GC-MSD). Based on the results, source profiles weredeveloped and discussed from the aspects of composition characteristics, potential tracers, BTEX (ben-zene, toluene, ethylbenzene and xylene) diagnostic ratios and chemical reactivity. Compared withvehicle exhausts and liquid fuels, the major components in refueling emissions of liquefied petroleumgas (LPG), gasoline and diesel were alkenes and alkanes. Oppositely, aromatics were the most abundantgroup in emissions from auto-painting, book binding and plastic producing. Three groups contributednearly equally in printing and commercial cooking emissions. Acetone in medical producing, chloroformand tetrachloroethylene in wet- and dry-cleaning, as well as TEX in plastic producing etc. were goodtracers for the respective sources. BTEX ratios showed that some but not all VOCs sources could bedistinguished by B/T, B/E and B/X ratios, while T/E, T/X and E/X ratios were not suitable as diagnosticindicators of different sources. The following reactivity analysis indicated that emissions from gasolinerefueling, commercial cooking, auto painting and plastic producing had high atmospheric reactivity, andshould be controlled emphatically to prevent ozone pollution, especially when there were large amountsof emissions for them.

© 2015 Elsevier Ltd. All rights reserved.

1. Introduction

As an important air pollutants in urban atmosphere, VOCs havebeing attracted more and more concerns in the world. They canchemically interact with oxides of nitrogen and sunlight to formground-level ozone, and even take part in the formation of haze,

lg@scies.com.cn (L. Chen).

which now is the most serious environmental problem in Chinawith frequent and large-scale occurrences in recent years (Anandet al., 2014; Durkee, 2014). More importantly, they also havedetrimental effects on human health by contributing to respiratoryillnesses, even being mutagenic or toxic to reproduction andharmful to the unborn (Anand et al., 2014).

Many studies indicated that chemical composition of VOCsemissions varied with different fuels or solvents, regions or coun-tries, and operating sectors or processes (Na et al., 2004; Liu et al.,2008; Zheng et al., 2013; Zhang et al., 2013). The diverse sources

Q. He et al. / Atmospheric Environment 115 (2015) 153e162154

and complex compositions of VOCs make it difficult to identifymajor emission sources and estimate the relative importance ofeach source to ambient VOCs concentrations, consequently hinderus from understanding the formation mechanism of secondarypollutions and devising effective control policies in urban areasseriously. Source profiles are the composition pattern of speciesemitted from each source category, and so are the fundamentalinformation for developing the speciated pollutants emission in-ventories used in air quality models, and identifying source con-tributions via receptor models (Zheng et al., 2009; Marmur et al.,2007). To make sure the accuracy of emission inventories andsource contributions, a unique, comprehensive and real-timeupdated “profile” (composition) is required for each source.

In present China, high population density (19.3% of the worldtotal in 2014, WPS, 2015) and fast growing economy beginning in1980s, have being accompanied by more and more pollutantsemission. For example, the total VOCs emission increased from13.1 Tg in 1995 to 20.1 Tg in 2005, and possibly would be 25.9 Tg in2020 as estimated by Wei et al. (2011). Scientists in China havebegun in recent years to measure local VOCs profiles for majoremission sources, such as vehicular exhaust, gasoline evaporation,coal burning, biomass burning, petrochemical industry and solventuse of painting and printing (Liu et al., 2008; Yuan et al., 2010; Guoet al., 2011; Zhang et al., 2013; Zheng et al., 2013; Wei et al., 2014;Wang et al., 2014). These studies have developed some domesticVOCs source profiles in China, but those profiles are far fromenough in covering all the VOCs sources such as LPG refueling,cleaning, plastic producing and book binding etc, or need moresupporting evidences, making it hard to establish accurate emissioninventories or apportion sources using CMB model in China (Wanget al., 2014a).

As a very large pollution source in China which is facing noto-rious air pollution, coal burning always attracts the most attention,but non-coal sources like fossil fuel evaporation (refueling, diurnaland hot-soak), solvent utilization and industrial processes cannotbe ignored. After America, China is the second-biggest consumer ofLPG and motor gasoline consuming 740.2 and 1615.9 thousandbarrels per day respectively in 2010 with the consumption stillincreasing (EIA, 2015a). During their handling and usage, largeamounts of VOCs are discharged. It was reported that approxi-mately 185,000 tons vehicular evaporative emissions were emittedin 2010 in China, among which refueling contributed 67% (Yanget al., 2015). For both solvent utilization and industrial processes,Wei et al. (2011) predicted they would become the highest twoVOCs contributors in 2020, emitting 37% (9.40 Tg) and 24% (5.92 Tg)of the total VOCs emissions, respectively, even if China's legislativestandards for VOCs emissions would be implemented effectively inthe future (Wei et al., 2011). To further enrich the database of VOCssource profiles in China, various non-coal emission sourcesincluding refueling of LPG and fossil oils, auto painting, printing,wet and dry cleaning, medicine and plastic producing, as well ascommercial cooking in a coastal region of China were investigated,with their potential tracers and BTEX ratios being discussed in ourstudy. Their atmospheric chemical reactivity was also evaluated togive more insights for targeted pollution control.

2. Methods

2.1. Sample collection

Different volumes of stainless steel canisters (Entech Instru-ment, Inc., Simi Valley, CA, USA) and Tedlar bags (Dalian Delin GasPacking Co., Ltd, China) pre-cleaned with high purity nitrogen andevacuated to vacuum were used for sampling. Whole-processsamplings were conducted for refueling, auto painting and

printing emissions to explore the general features for them. For theemissions of medical producing, plastic processing and commercialcooking, instantaneous sampling above the exhaust ducts wasconducted by the aid of a pump connected to the sampling bags.After sampling, all the samples were delivered to the laboratory forchemical analysis within aweek. Detailed information is as follows:

2.1.1. Re-fueling of LPGIn China, a large proportion of residents living in rural places and

city villages used the tanked-LPG for cooking instead of coal forconvenience and light pollution. The tanks can be used repeatedlyby re-inflating. For each sample, refueling emissions were collectedfrom the launch to the stop of inflating operation to a civilian gastank (capacity, 13.5 kg) using a 400-mL canister in an inflatingworkshop (n ¼ 10; T, 24.5 �C). Before collecting, the connectingjoint between gas inflating and releasing nozzle was wrapped toform a small enclosed space. One side of a tubule was also wrappedin it while the other side was connected to a sampling canister. Theaverage sampling duration was 9.64 ± 0.64 s.

2.1.2. Re-fueling of gasoline and dieselOil refueling operations include transferring fuel from bulk

storage tanks into transport vehicles like road tankers, road tankersinto service station tanks and service station tanks into vehicletanks. During filling of tanks, vapor evolves as a result ofdisplacement of headspace vapor by the incoming oil, splashingand turbulence. In our study, vapors from storage tanks and vehicletanks during refueling with gasoline and diesel were collected into400-mL canisters through inserting sampling intake-tubules intothe upper space of the tanks without touching the liquid surface.For gasoline, the vehicles investigated included cars and motorcy-cles. The sampling number for each refueling source was 10. Rushhours (9:00e10:00 a.m.) were chosen for sampling. The averagesampling duration was 2.57 ± 0.39 min for storage tanks, and36 ± 9.61 s for vehicle tanks. The ambient temperature during thesampling period was 28.3 �C with no strong wind.

2.1.3. Auto painting (automobile painting)Painting is an important work in car-related industries. In an

auto-painting workshop of a typical garage where gasoline-basedpaints were still the common used paints, 12 canister sampleswere collected in three different corners far from the door (4samples for each corner) while 8 bag samples (2.5-L) were collectedfrom the exhaust duct of the auto-painting workshop. There arethree most important processes in painting, namely primer paint-ing, finish coat painting and paint-drying with different paintsbeing used (Yuan et al., 2010).With 1L paints for 5m2 car surface onaverage, sampling inside the workshop lasted 108 min covering allthe 3 steps of painting with the indoor temperature being 24.9 �C.To avoid influences from other emission sources, all windows in thesampling roomwere closed and canisters were held near the VOCssources.

2.1.4. PrintingSampling was conducted in a comparatively large printing fac-

tory that mainly printed books, magazines, newspapers, andpackaging materials. The press equipments in the factory were alloffset lithographic printing machines. 10 canister samples (2-L)were collected at different sites of the printing room (total areawas4000 square feet), and another 10 were collected in binding room(total area was 5000 square feet) when all the press equipmentswere operating. The sampling duration was 1 h, during which 20thousand papers were printed. As the sampling for auto-painting,windows in the sampling rooms were also closed. The averagetemperature in the printing factory was 23.2 �C.

Q. He et al. / Atmospheric Environment 115 (2015) 153e162 155

2.1.5. Commercial cleaningVOCs emission from commercial cleaning is due to the use of

detergents. The major components in the wet-cleaning detergentswere the mix of chloroform and tetrachloroethylene, while in thedry-cleaning detergents tetrachloroethylene was the uniquecomponent. To explore the VOCs emission profiles of wet- and dry-cleaning, the valves of sampling canisters (400-mL) were openedand put into washing machines promptly just after they stoppedworking, and then the lids were immediately closed until thesampling finished. The samplings were conducted during workhours, and lasted 15 min for wet cleaning and 10 min for drycleaning. The average environment temperature during samplingwas 25 �C.

2.1.6. Medicine producingThe pharmaceutical factory investigated in this study mainly

produced western medicine including oral drugs, parenteral drugs,inhalant and topical drugs, etc. 8, 10 and 10 samples of 500 ml bagswere collected above the exhaust ducts of manufacturing work-shops, wastewater treatment workshops and centrifugal work-shops respectively during several intermittent exhaust stages withone sample covering one exhaust stage. The average temperaturenear exhaust ducts was 25.0 �C.

2.1.7. Plastic processingPlastics processing, is also commonly referred to as polymer

processing, involves molding, forming, shaping, or altering plasticresins or plastic materials. All of the operations could give rise toVOCs emissions, resulting mainly from the volatilization of freemonomer or solvent in the primary polymer blend during pro-cessing, secondary process materials, such as blowing agents, ad-ditives, and lubricants (mold release compounds) and byproductsformed by chemical reactions or formed during heating of resins. 15bag samples (500-mL) were collected above the exhaust ductsduring the melting processes of propene polymer (PP) and polyethylene (PE) respectively in a plastic processing factory. Theaverage temperature near exhaust ducts was 34.7 �C.

2.1.8. Commercial cooking8 and 7 bag samples (500-mL) were collected above the exhaust

ducts of a Chinese restaurant and a barbecue restaurant respec-tively during peak-hour business (at noon), and the average tem-peratures near their exhaust ducts were 35.6 and 37.3 �Crespectively. The major cooking styles in Chinese restaurant weredeep-fry, stir-fry, steam and boil, all of which were done in a uniquekitchen with LPG being used as the major cooking fuel, while thestyles in barbecue restaurant were roast and fumigate on electricoven with one exhaust pipe hanging above each dinner table andthe emissions converging finally in the ducts.

2.2. Chemical analysis and quality assurance

VOCs samples were concentrated in the Model 7100 pre-concentrator (Entech Instruments Inc., USA), and then transferredinto a gas chromatography-mass selective detector (GC-MSD/FID,Agilent 7890A/5975C, USA). A HP-1 capillary column(60 m � 0.32 mm � 1.0 mm, Agilent Technologies, USA) was firstused in this system to separate the mixture, which was then splitinto two ways: one is to a PLOT-Q column(30m� 0.32mm� 2.0mm, Agilent Technologies, USA) followed byFID for detecting C2eC3 hydrocarbons; another is to a80 cm� 0.10mm I.D stainless steel line followed byMSD for C4eC12hydrocarbons. The MSD was used in selected ion monitoring (SIM)mode and the ionization method was electron impacting (EI).Target compounds were identified based on their retention times

and mass spectra, and quantified by multi-point external calibra-tion method with TO-14 (39 compounds) and TO-15 (25 com-pounds) standard mixtures from Spectra Gases Inc., USA. To avoidmisjudgment, the retention times of VOCs species were determinedalso by MSD with SCAN mode for qualitative accuracy. The detaileddescription can be found elsewhere (Zhang et al., 2012). Themethod detection limits of target compounds in present studywereall below 0.1 mg/m3.

2.3. Data analysis

In each sample, 62 VOCs species were measured as shown inTable S1. VOCs source profiles were developed by using averagepercentage of mass known as weight fraction. For each source, theaverage constant OH loss rate (KOH) and the maximum incrementalreactivity (MIR) (sign as k-avg and MIR-avg) of VOCs, and therelative contribution from each species to the total VOCs reactivity(OH loss rate (LOH) and the ozone formation potential (OFP)) wereestimated according to themethods reported byWang et al. (2014).

3. Results and discussion

3.1. Refueling emissions

Vehicle refueling, emitting 124,000 tons VOCs in 2010, hasbecome an important source of VOCs related with fossil fuel usageexcept for vehicle emissions (Yang et al., 2015). VOCs source profilesfor LPG, gasoline and diesel refueling investigated in the currentstudy were shown in Fig. 1. Alkanes and alkenes were in highproportion (>86%) in all of the refueling emissions, with higherhydrocarbons based in gasoline (>C4) and diesel (>C4), while lightalkanes (<C5) based in LPG. Some C4 compounds like isobutane,cis-2-butene and isopentane could be emitted from any of them(Table S2). Aromatics, the main components in fuel (Zhang et al.,2013), took up high proportions in gasoline and diesel vehicle ex-hausts (Liu et al., 2008; Guo et al., 2011; Schauer et al., 1999) butlittle in headspace vapors (Zhang et al., 2013) and refueling emis-sions (Our study and Zhang et al., 2013) for their relative low vaporpressure. In gasoline refueling emissions, the percentage of alkanes(29.6e35.56%) from our study were lower than that of 55.33e65.8%reported by Zhang et al. (2013), which was opposite for alkenes andaromatics.

Different filling rates and different ways of splashing and tur-bulence might lead to the small variations about the VOCscomposition among the refueling emissions of gasoline from stor-age tanks, as well as car and motorcycle tanks with different innerstructures (Fig. 1B), while different oil composition was the majorreason for the differences among the emissions from industrial,low-sulfur and vehicle-used diesels (Fig. 1C). Except for mutablecomposition of the oils in the market of China, different samplingmethods also contributed to the difference. Our data gave theaverage situation during the whole refueling process, comparedwith the special situation when the vacuum assisted recoverysystems were turned on and turned off. We inserted samplingintake-tubules connected with canisters into the upper space of thetank, but Zhang et al. (2013) just collected the vapors about 2 cmabove the car tank vents.

Propane and butanes werewidely used as tracers of LPG leakagein some countries like USA and Korea etc., where only propane or n-butane or they both dominated the total emission (Ayoko et al.,2014; Na et al., 2004), while our study found that the composi-tion of LPG refueling emissions in China was more complex. Itincluded 6 major species: n-butane, isobutane, acetylene, ethane,propane and propylene (Table S2). Among them acetylene, ethaneand propylene were characteristic compounds of LPG which could

Fig. 1. VOCs source profiles of the emissions from refueling of LPG (A), gasoline (B), and diesel (C). The ID number of each species was the same as that of Table S1 (No. 1e59). Hollowpillars represented the data from our study, while colorized dots showed the data from other reported studies. (For interpretation of the references to color in this figure legend, thereader is referred to the web version of this article.)

Q. He et al. / Atmospheric Environment 115 (2015) 153e162156

distinguish it from the refueling vapors of gasoline and diesel(compared Fig. 1A with others). Combination of abundant butenes,isopentane, 2-methylpentane and MTBE could be used to differ-entiate gasoline and diesel refueling emissions from other sources,while combination of high percentages of isopentane, toluene andMTBE could be used to distinguish gasoline vehicle exhausts fromother sources concerned in our study (compared Fig. 1B withothers). Some heavier alkanes, such as n-heptane, n-undecane andmethylcyclohexane, which were relatively abundant in diesel,especially in vehicle-used diesel but low in gasoline, might be taken

as indicators for distinguishing the diesel refueling emissions fromgasoline (Table S2). AlthoughMTBEwas often used as a tracer of thesources relative to gasoline, its role in diesel refueling emissionscould not be ignored (Fig. 1C).

3.2. Solvent use emissions

Common activities involving solvent use in urban cities mainlyinclude paint application, printing processes, wet and dry cleaning,etc. In China, small-sized workshops offering the above services are

Q. He et al. / Atmospheric Environment 115 (2015) 153e162 157

not only widespread but popular among people for their conve-nience and low expense in many neighborhoods. They alwaysdirectly discharge their organic exhausts into the air, withoutexhaust collection and treatment systems. In our study, represen-tative individual workshops were investigated as shown in Fig. 2 toprovide more knowledge about them. Those samples collected in-side the manufacturing workshops were classified as “fugitiveemission” samples, while those collected from the chimneys wereclassified as “stack emission” samples.

3.2.1. Auto paintingIn fugitive emissions, BTEX were found to be the main trace

species, with m/p-xylene accounting for the largest percentage,followed by ethyl-benzene, o-xylene, toluene (Fig. 2A; Table S3).The results were consistent with other reported analyses (Yuanet al., 2010; Wang et al., 2014), which also found the same aro-matic tracers in paint application emissions (Fig. 2A).

3.2.2. PrintingStudies in VOCs profiles of printing source from this study and

other reported data (Yuan et al., 2010; Wang et al., 2014) all indi-cated that alkanes were the most important species in the emis-sions followed by aromatics, but different compositioncharacteristics were observed (Fig. 2B). In our study, vinyl acetatewas only detected in printing emissions, so it should be animportant tracer for this source. And toluene was found to be themost abundant species in the printing emissions contributing12.74% of the TVOCs, MTBE, m,p-xylene and isopentane, followed itand contributed 6e8% respectively (Table S3), while the contribu-tion of each of the rest 80% compounds was only 0.1e3% by mass(data not shown). However, n-decane (16.86%), n-nonane (14.84%)and n-undecane (12.96%) were found to be the abundant species ina printing factory by Yuan et al. (2010), while ethane (17.4), n-

Fig. 2. VOCs source profiles of the emissions from auto painting (A) and printing (B). The Irepresented the data from this study, while colorized dots showed the data from other reportis referred to the web version of this article.)

pentane (13.1%), m,p-xylene (10.8%) and ethylbenzene (10.1%) etc.were found by Wang et al. (2014a,b) in mixed printing exhausts. Itwas learned that printing machines were often cleaned usinggasoline, which could explain the high levels of isopentane andMTBE. The binding of printed materials using adhesives alsoemitted VOCs (Fig. 2B), but the average concentration of fugitiveemissions in the binding room (1.41mg/m3) was significantly lowerthan that in the printing room (23.04 mg/m3), and m,p-xylene andethyl-benzene were the major species accounting for 46.72% and14.52% of the total VOCs (Table S3).

3.2.3. Commercial wet- and dry-cleaningChloroform and tetrachloroethylene were found only in the

emissions of dry- andwet-cleaning, andwere good tracers for theseactivities. Chloroform and tetrachloroethylene accounted for 78%and 22% respectively by mass in wet-cleaning. Also, tetrachloro-ethylene was virtually the only component (z100% of the total)emitted fromdry cleaning process. The results were consistentwiththose conducted in drying cleaning buildings or facilities(McDermott et al., 2005; Chiappini et al., 2009), residential homes(Kinney et al., 2002) and buildings near drying-cleaning facilities(Kwon et al., 2006).

3.3. Industrial and commercial emissions

3.3.1. Pharmaceutical industry14 VOCs species were detected in the samples from the medi-

cine producing industry. Acetone, a commonly used solvent inpharmaceutical industry, not only accounted for high percentage(>50%) in various pharmaceutical processes (our study and Heet al., 2012) including production, wastewater treatment andcentrifugation, but also was unique compared with other emissionsources involved in our study. In emissions from production lines

D number of each species was the same as that of Table S1 (No. 1e59). Hollow pillarsed studies. (For interpretation of the references to color in this figure legend, the reader

Q. He et al. / Atmospheric Environment 115 (2015) 153e162158

and centrifugation, the second abundant individual species was n-pentane, accounting for 16.04% and 12.90% respectively, while inemissions from wastewater treatment, the second was tolueneaccounting for up to 43.66%. The rest VOCs species from the threesampling sites were n-heptane and other 11 benzene homologues(m, p-xylene, ethyl-benzene, o-xylene, benzene, 1, 2, 4-trimethylbenzene, o-ethyltoluene, m-ethyltoluene, 1, 3, 5-trimethyl-benzene, p-ethyltoluene and n-propylbenzene)(Table S4).

3.3.2. Plastic processingThe previous studies on the PE/PP plastic waste recycling found

that four toxic monocyclic aromatic hydrocarbons (MAHs) (m,p-xylene, ethyl-benzene, o-Xylene and toluene) had the highestoverall concentrations in the volatile compounds emitted from themelting and powdering processes (Tsai et al., 2009; Huang et al.,2013). Also, our study found that the four species were also themain components of the VOCs exhausted by plastic processing,accounting for more than 98% of the total either for PP producing orfor PE producing. However, the composition characteristics of theemissions from plastic recycling and producing were entirelydifferent. Toluene only contributed 5.66% and 2.11% to the totalVOCs emissions from PP and PE producing (our study), butaccounted for about 40% of the total hydrocarbons emitting frommelting and powering process (Tsai et al., 2009; Huang et al., 2013).And it was just opposite for m, p-xylene, ethyl-benzene and o-Xylene (Table S4).

3.3.3. Commercial cookingEmission of VOCs in cooking events have drawn considerable

concerns in recent years (Huang et al., 2011; Mugica et al., 2001) inChina because of the characteristic stir-frying cooking process(stove and food) with oil being heated to a very high temperature.In our Chinese restaurant samples, alkanes and aromatics accoun-ted for 50.85% and 35.39% of the TVOCs respectively, while alkenesaccounted for 12.51%. Among the species, n-pentane (21.69%) wasthe most abundant component in Chinese restaurant exhaust, fol-lowed by styrene (11.98%) and m, p-xylene (11.10%) (Table S4). Inour barbecue restaurant samples, the contribution of alkanes, ar-omatics and alkenes were 41.75%, 28.35% and 27.23% separately,while contributions from n-pentane, ethylene and styrene were21.97%, 15.61% and 9.87% respectively (Table S4). Direct evaporationlosses of cooking fuels, incomplete combustion of meat and grease,cooking methods and charcoal used etc. all could influence theVOCs composition of cooking emissions and make it relativelymutable in different studies (Huang et al., 2011; Mugica et al.,2001).

3.4. BTEX ratios

According to the analysis of VOCs composition characteristics ofeach source emissions in our study, the species that were unique orpresent in high concentrations in a certain emission source wereselected as possible tracers for it (Table 1). Because the chemicalcomposition of printing and cooking emissions was variable, thetracers given in our study need to be confirmed by more studies.Some sources emitting similar VOCs species led to the inevitableoverlap, which can complicate the application of CMB models. Thediagnostic ratios of paired VOCs species well correlated in ambientmeasurements have been demonstrated to be a good way todifferentiate the overlapped sources, such as benzene to tolueneratio (B/T), around 0.5 (wt/wt) being reported to be a characteristicof vehicular emissions, while >1 the characteristic of biomassburning (Barletta et al., 2005; Tan et al., 2012). As common speciesin the real environment and always being discharged from the

same source, the paired ratios of BTEX received more attention, andtherefore were analyzed for each source using the data from ourstudy and previous works (Zhang et al., 2013; Wang et al., 2014;Yuan et al., 2010).

As shown in Table 2, B/T ratios ranged 0.3e1.11 in refueling ofgasoline and diesel emissions, 0.01e0.3 in solvent use emissions,and 0.005e0.016 in industrial activity emissions, while >1 wasreported to be the characteristic of biomass burning (Barletta et al.,2005; Tan et al., 2012). Little overlap among these sources indicatedthat B/T ratio could be a good marker to distinguish them whenconducting source apportionment of atmospheric VOCs using re-ceptor models. Some challenges were also found for the applicationof B/T ratio. The LPG refueling emission had a B/T ratio of 1.7 as thebiomass burning (>1). B/T ratios in refueling emissions of gasolineand diesel (Table 2) were similar with the ratios found in vehicleemissions (0.33e0.67 was proposed to be the typical range of B/Tratios for traffic dominated source of emission, Shi et al. (2015)). Inaddition, the average B/T ratios of 0.60 and 0.32 were also found inthe emissions from coke production and thermal power plantrespectively (Shi et al., 2015). Barletta et al. (2008) proposed spe-cific B/T ratios of <0.20 as an indicator for air samples which werestrongly affected by industrial emissions. Furthermore, auto-painting, printing, medicine producing, plastic producing andcommercial cooking all fit this characteristic at some extent andoverlapped with each other (Table 2), making it hard to differen-tiate them well in detail.

Ratios of B/E (>2.38) and B/X (>0.35) for refueling emissions ofgasoline and diesel were much higher than their counterparts inother emissions from painting, printing, medicine producing andcommercial cooking, but overlapped with the ratios for that of coalburning (about 3 for B/X), coke production (about 3 for both B/E andB/X) and vehicle exhausts (about 2.5 for B/E and 0.4 for B/X) etc.when calculated using the average data reported by other re-searchers (Shi et al., 2015; Liu et al., 2008; Na et al., 2004) (Table 2).Therefore, it will be better to have a general knowledge beforeusing them to explore the sources. There were more overlaps for T/E, T/X and E/X ratios, so they are not suitable as indicators ofdifferent sources.

3.5. Reactivity characteristics

VOCs could influence the ozone formation and the atmosphericoxidizing capacity after being emitted into the atmosphere(Lelieveld et al., 2008), so the sources or the compounds in a certainsource having high atmospheric chemical reactivity should begiven priority when devising control programs. In our study, the k-avg and MIR-avg values representing the average VOCs reactivity ofeach source were given in Table 3. It can be seen that the k-avg ofVOCs from gasoline refueling were the largest, which were fol-lowed by that from commercial cooking, auto painting and plasticproducing. The k-avg for VOCs from gasoline refueling sampled inthis studywas similar to that of m-ethyltoluene (Atkinson and Arey,2003), while the k-avg from commercial cooking, auto painting andplastic producing was a little higher than that of p-xylene. There-fore, those sources will serve as important reactive sources in theregions unless their emission amounts are very low. VOCs exhaustfrom the medicine producing had the smallest k-avg value whichwas close to that of n-butane, indicating it might have low effect onthe atmospheric chemical reactivity.

The sources with high k-avg also had highMIR-avg. In our study,the MIR-avg of VOCs from the plastic producing was the highest,followed by that of auto painting, book binding, gasoline refueling,printing, and commercial cooking in the decreasing order. Coin-ciding with the k-avg, VOCs from the medicine producing exhausthad the lowest MIR-avg. The MIR-avg of VOCs emitted from the

Table 1The potential VOCs tracers for various emission sources in China.

Type Source Potential tracers

Fuel refueling LPG Butanes, propane, acetylene, ethane, propyleneGasoline MTBE, butenes, isopentane, 2-methypentane,Diesel MTBE, butenes, isopentane n-undecane, methylcyclohexane, n-heptane

Solvent use Auto painting m,p-Xylene, ethyl-benzene, o-xylene, toluenePrinting Vinyl acetate,Wet cleaning Chloroform, tetrachloroethylene,Dry cleaning Tetrachloroethylene

Industrial activities Medicine producing AcetonePlastic producing Xylenes, ethyl-benzene, tolueneCommercial cooking n-Pentane, styrene, m,p-xylene, ethylene

Table 2The value of BTEX ratios for different emission sources.

Type Source B:T B:E B:X T:E T:X E:X

Fuel refueling LPGa 1.70 e 1.94 e 1.14 0.00Gasoline to STa 0.70 14.39 5.78 20.58 8.26 0.40Gasoline to motorcyclesa 0.30 16.49 0.35 55.88 1.18 0.02Gasoline to carsa 0.35 3.51 1.11 9.99 3.15 0.32Gasoline to tank truckb 0.31 5.92 1.97 18.92 6.31 0.33Gasoline to carsb 0.50 4.74 3.03 9.57 6.11 0.64Industrial diesel to STa 0.59 2.64 1.01 4.44 1.70 0.38Low-sulfur diesel to STa 1.11 12.28 4.41 11.04 3.97 0.36Diesel to car to STa 0.42 2.38 0.72 5.71 1.73 0.30

Solvent use Auto painting (fugitive)a 0.02 0.01 0.00 0.23 0.12 0.53Auto-paint spraying (fugitive)c 0.27 0.027 0.01 0.08 0.03 0.36Auto-paint drying (fugitive)c 0.04 0.02 0.01 0.48 0.18 0.37Auto paintsd 0.01 0.03 0.01 2.31 0.61 0.27Printing (fugitive)a 0.08 0.28 0.09 3.58 1.17 0.33Book binding (fugitive)a 0.18 0.04 0.01 0.21 0.06 0.27Print (stack)d 0.13 0.05 0.03 0.39 0.23 0.60Printing factory (fugitive)c 0.15 0.21 0.08 1.38 0.51 0.37

Industrial and commercial activities Medicine producing lines 0.08 0.25 0.09 2.97 1.11 0.37Medical wastewater treatment 0.01 0.30 0.16 53.20 27.93 0.53Medicine centrifugation 0.08 0.34 0.24 4.04 2.91 0.72PP producing 0.01 0.00 0.00 0.25 0.08 0.33PE producing 0.02 0.00 0.00 0.09 0.03 0.30Chinese restaurant cooking 0.12 0.13 0.03 1.03 0.28 0.27Barbecue restaurant cooking 0.32 0.47 0.13 1.48 0.42 0.28

BTEX, benzene, toluene, ethylbenzene and xylene; LPG, liquefied petroleum gas; ST, storage tanks.a This study.b Zhang et al., 2013.c Wang et al., 2014.d Yuan et al., 2010.

Table 3Comparison of the k-avg and MIR-avg among different emission sources.

Type Source k-avg (10�12 cm3/molecule/s) MIR-avg (g O3/g VOC)

Fuel refueling LPG 7.42 2.03Gasoline to ST 21.23 5.14Gasoline to motorcycles 19.16 4.79Gasoline to cars 22.11 5.37Industrial diesel to ST 85.11 2.46Low-sulfur diesel to ST 14.44 3.70Diesel to cars 11.65 1.88

Solvent use Auto painting (fugitive) 14.26 5.91Printing (fugitive) 10.65 3.45Book binding (fugitive) 13.75 5.35

Industrial activities Medicine producing lines 2.37 1.34Medical wastewater treatment 2.85 2.21Medicine centrifugation 1.41 0.97PP producing 14.02 6.21PE producing 14.74 6.42

Commercial activities Chinese restaurant cooking 16.51 3.34Barbecue restaurant cooking 15.12 2.93

ST, storage tanks.

Q. He et al. / Atmospheric Environment 115 (2015) 153e162 159

Q. He et al. / Atmospheric Environment 115 (2015) 153e162160

plastic producing, auto paint and book binding in this study werehigher than that from vehicle exhaust, biomass burning and coalcombustion (Yuan et al., 2010). For painting and printing sources,the MIR-avgs of their emissions obtained in our study were close tothe levels found in other studies (Wang et al., 2014; Yuan et al.,2010).

To provide more information to the targeted control of ambientozone pollution and atmospheric oxidation, the top 10 VOCs spe-cies that contributed the most to the total LOH and OFP of eachsource were listed in Table S5e10. For VOCs in emissions fromrefueling of LPG and gasoline, auto painting, book binding, medi-cine producing, plastic producing and commercial cooking, beyond75% of their reactivity or LOH and OFPs was explained by the top 10

Fig. 3. VOCs source profiles from medicine producing (A), plastic producing (B) and comme1e59).

species; On the contrary, the top 10 species just shared 60e70% ofthe total reactivity for emissions from diesel refueling and printing.

4. Summary

The composition characteristics and BTEX ratios of VOCs emis-sions from several major sources, including fuel refueling, autopainting, printing, medicine producing, plastic producing, com-mercial cooking, wet and dry cleaning in China were experimen-tally determined and compared with other studies. Becausetrichloroethylene, tetrachloroethylene and chloroform (No. 60e62in Table S1) were only emitted from dry- and wet-cleaning, andwere the unique species emitted from this source, they were

rcial cooking (C). The ID number of each species was the same as that of Table S1 (No.

Q. He et al. / Atmospheric Environment 115 (2015) 153e162 161

excluded from the profiles shown in Figs. 1e3. Many factors couldinfluence the chemical composition for one source, like differentfuels and solvent composition, different technical processes andsampling methods etc. For some sources like gasoline refueling,auto painting, medicine producing, plastic producing, wet- anddry-cleaning, the species composition and tracers of their emis-sions were almost the same with other studies, so their profilesshould be applicable in CMB model, and be taken as references tomatch with the results from PMF receptor model. For the printingand cooking emissions, their source profiles were variable and notfit to be used in receptor models for source apportionment.

B/T was a good marker to justify the pollution characteristics,and also to differentiate the traffic and combustion emissionsources in a mixed environment polluted by local emissions. Ourstudy further found that B/T ratio could be a good marker todistinguish the emissions from refueling of gasoline and diesel,solvent use and industrial activities. B/E and B/X ratios can also beused to differentiate some sources. Generally speaking, for a mixedenvironment, the more simplex the source composition was, thecloser the ratio in that environment was to the ratio of the majorsource. On the other hand, our study also proposed some chal-lenges about their application, which should be taken intoconsideration in actual use.

Ozone pollution has become a global problem, and VOCs areimportant precursors for the formation of ozone through reactingwith hydroxyl radical $OH. According to our study, the emissions ofgasoline refueling, commercial cooking, auto painting, book bind-ing and plastic producing were highly reactive, and could increaseatmospheric loss of $OH and O3 formation. Therefore, they shouldbe listed as precedence-controlled sources in solving O3 pollutionespecially when there are large amounts of emissions.

China is the world's top coal producer, consumer and importer,and accounts for about half of the global coal consumption. Ac-cording to the statistic data of EIA (2015b), coal supply constitutedthe vast majority (69%) of China's total energy consumption in 2011(EIA, 2015b). In addition to industrial applications (mainly for in-dustrial boilers and furnaces), coal is used throughout the countryin power plant and household stoves for cooking and heatingpurposes. Coking plants are also very common around China(Barletta et al., 2005). Therefore, building up coal-related VOCsprofiles should be our next goals.

Acknowledgments

This study was supported by the funding of the National Natural Science Foundation of China (No. 41172316), the Key Project of The National Ministry of Education of China (No. 211026), Shanxi Province Science Foundation for Youths (No. 2011021025-2, 2015021059), the Research Project Supported by Shanxi Scholar-ship Council of China (2011080) and Doctoral Scientific Research Foundation of TYUST (20132018).

Appendix A. Supplementary data

Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.atmosenv.2015.05.066.

References

Anand, S.S., Philip, B.K., Mehendale, H.M., 2014. Volatile organic compounds. In:Philip, W. (Ed.), Encyclopedia of Toxicology, third ed. Elsevier Inc., pp. 967e970.

Atkinson, R., Arey, J., 2003. Atmospheric degradation of volatile organic compounds.Chem. Rev. 103, 4605e4638.

Ayoko, G.A., Singh, A., Lim, M.C.H., Ristosvski, Z.D., Jayaratne, E.R., Morawska, L.,

King, G., Christensen, E., 2014. Characterization of VOCs from LPG and unleadedpetroleum fuelled passenger cars. Fuel 115, 636e643.

Barletta, B., Meinardi, S., Rowland, F.S., Chan, C.Y., Wang, X., Zou, S., Chan, L.Y.,Blake, D.R., 2005. Volatile organic compounds in 43 Chinese cities. Atmos.Environ. 39, 5979e5990.

Barletta, B., Meinardi, S., Simpson, I.J., Zou, S., Sherwood Rowland, F., Bslake, D.R.,2008. Ambient mixing ratios of nonmethane hydrocarbons (NMHCs) in twomajor urban centers of the Pearl River Delta (PRD) region: Guangzhou andDongguan. Atmos. Environ. 42, 4393e4408.

Chiappini, L., Delery, L., Leoz, E., Brouard, B., Fagault, Y., 2009. A first Frenchassessment of population exposure to tetrachloroethylene from small dry-cleaning facilities. Indoor Air 19, 226e233.

Durkee, J.B., 2014. Appendix B1-chemistry of Atmospheric Reactions of VOCsLeading to Smog. In: Cleaning with Solvents: Science and Technology. ElsevierInc., pp. 547e556

EIA (Energy Information Administration), 2015a. http://www.eia.gov/cfapps/ipdbproject/iedindex3.cfm?tid¼5&pid¼62&aid¼2&cid¼regions&syid¼2008&eyid¼2012&unit¼TBPD(accessed 27.02.15.).

EIA (Energy Information Administration), 2015b. http://www.eia.gov/countries/cab.cfm?fips¼ch (accessed 13.12.14.).

Guo, H., Zou, S.C., Tsai, W.Y., Chan, L.Y., Blake, D.R., 2011. Emission characteristics ofnonmethane hydrocarbons from private cars and taxis at different drivingspeeds in Hong Kong. Atmos. Environ. 45, 2711e2721.

He, H., Wang, Z., Xu, M., He, Z., Song, S., 2012. Studies on the emission character-istics and countermeasures of VOCs from pharmaceutical industrydbased onZhejiang Province. China Environ. Sci. 32, 2271e2277.

Huang, D.Y., Zhou, S.G., Wei, H., Wei, F.F., Liang, T., 2013. Pollution characteristics ofvolatile organic compounds, polycyclic aromatic hydrocarbons and phthalateesters emitted from plastic wastes recycling granulation plants in XingtanTown, South China. Atmos. Environ. 71, 327e334.

Huang, Y., Ho, S.S.H., Ho, K.F., Lee, S.C., Yu, J.Z., Louie, P.K.K., 2011. Characteristics andhealth impacts of VOCs and carbonyls associated with residential cooking ac-tivities in Hong Kong. J. Hazard. Mater. 186, 344e351.

Kinney, P.L., Chillrud, S.N., Ramstrom, S., Ross, J., Spengler, J.D., 2002. Exposures tomultiple air toxics in New York City. Environ. Health Perspect. 110 (Suppl. 4),539e546.

Kwon, J., Weisel, C.P., Turpin, B.J., Zhang, J.F., Korn, L.R., Morandi, M.T., Stock, T.H.,Colome, S., 2006. Source proximity and outdoor-residential VOC concentra-tions: results from the RIOPA study. Environ. Sci. Technol. 40, 4074e4082.

Lelieveld, J., Butler, T.M., Crowley, J.N., Dillon, T.J., Fischer, H., Ganzeveld, L.,Harder, H., Lawrence, M.G., Martinez, M., Taraborrelli, D., Williams, J., 2008.Atmospheric oxidation capacity sustained by a tropical forest. Nature 452,737e740.

Liu, Y., Shao, M., Fu, L., Lu, S., Zeng, L., Tang, D., 2008. Source profiles of volatileorganic compounds (VOCs) measured in China: part I. Atmos. Environ. 42,6247e6260.

Marmur, A., Mulholland, J.A., Russell, A.G., 2007. Optimized variable source-profileapproach for source apportionment. Atmos. Environ. 41, 493e505.

McDermott, M.J., Mazor, K.A., Shost, S.J., Narang, R.S., Aldous, K.M., Storm, J.E., 2005.Tetrachloroethylene (PCE, Perc) levels in residential dry cleaner buildings indiverse communities in New York City. Environ. Health Perspect. 113,1336e1343.

Mugica, V., Vega, E., Chow, J., Reyes, E., S�anchez, G., Arriaga, J., Egami, R., Watson, J.,2001. Speciated non-methane organic compounds emissions from food cookingin Mexico. Atmos. Environ. 35, 1729e1734.

Na, K., Kim, Y.P., Moon, I., Moon, K.C., 2004. Chemical composition of VOC majoremission sources in the Seoul atmosphere. Chemosphere 55, 585e594.

Schauer, J.J., Kleeman, M.J., Cass, G.R., Simoneit, B.R.T., 1999. Measurement ofemissions from air pollution sources. 2. C2 through C30 organic compoundsfrom medium duty diesel trucks. Environ. Sci. Technol. 33, 1578e1587.

Shi, J., Deng, H., Bai, Z., Kong, S., Wang, X., Hao, J., Han, X., Ning, P., 2015. Emissionand profile characteristic of volatile organic compounds emitted from cokeproduction, iron smelt, heating station and power plant in Liaoning Province,China. Sci. Total Environ. 515e516, 101e108.

Tan, J.H., Guo, S.J., Ma, Y.L., Yang, F.M., He, K.B., Yu, Y.C., Wang, J.W., Shi, Z.B.,Chen, G.C., 2012. Non-methane hydrocarbons and their ozone formation po-tentials in Foshan, China. Aerosol Air Qual. Res. 12, 387e398.

Tsai, C.J., Chen, M.L., Chang, K.F., Chang, F.K., Mao, I.F., 2009. The pollution charac-teristics of odor, volatile organochlorinated compounds and polycyclic aromatichydrocarbons emitted from plastic waste recycling plants. Chemosphere 74,1104e1110.

Wang, H., Qiao, Y., Chen, C., Lu, J., Dai, H., Qiao, L., Lou, S., Huang, C., Li, L., Jing, S.,Wu, J., 2014a. Source profiles and chemical reactivity of volatile organic com-pounds from solvent use in Shanghai, China. Aerosol Air Qual. Res. 14, 301e310.

Wang, M., Shao, M., Chen, W., Yuan, B., Lu, S., Zhang, Q., Zeng, L., Wang, Q., 2014b.A temporally and spatially resolved validation of emission inventories bymeasurements of ambient volatile organic compounds in Beijing, China. Atmos.Chem. Phys. 14, 5871e5891.

World Population Statistics, 2015. http://www.worldpopulationstatistics.com/population-of-china-2014/ (accessed 13.12.15.).

Wei, W., Cheng, S., Li, G., Wang, G., Wang, H., 2014. Characteristics of volatile organiccompounds (VOCs) emitted from a petroleum refinery in Beijing, China. Atmos.Environ. 89, 358e366.

Wei, W., Wang, S., Hao, J., Cheng, S., 2011. Projection of anthropogenic volatile

Q. He et al. / Atmospheric Environment 115 (2015) 153e162162

organic compounds (VOCs) emissions in China for the period 2010-2020.Atmos. Environ. 45, 6863e6871.

Yang, X., Liu, H., Cui, H., Man, H., Fu, M., Hao, J., He, K., 2015. Vehicular volatileorganic compound losses due to refueling and diurnal process in China:2010e2050. J. Environ. Sci. 33, 88e96.

Yuan, B., Shao, M., Lu, S., Wang, B., 2010. Source profiles of volatile organic com-pounds associated with solvent use in Beijing, China. Atmos. Environ. 44,1919e1926.

Zhang, Y., Li, C., Wang, X., Guo, H., Feng, Y., Chen, J., 2012. Rush-hour aromatic andchlorinated hydrocarbons in selected subway stations of Shanghai, China.J. Environ. Sci. 24, 131e141.

Zhang, Y., Wang, X., Zhang, Z., Lü, S., Shao, M., Lee, F.S.C., Yu, J., 2013. Species profilesand normalized reactivity of volatile organic compounds from gasoline evap-oration in China. Atmos. Environ. 79, 110e118.

Zheng, J.Y., Shao, M., Che, W., Zhang, L., Zhong, L., Zhang, Y., Streets, D., 2009.Speciated VOC emission inventory and spatial pattern of ozone formation po-tentials in the Pearl River Delta, China. Environ. Sci. Technol. 43, 8580e8586.

Zheng, J., Yu, Y., Mo, Z., Zhang, Z., Wang, X., Yin, S., Peng, K., Yang, Y., Feng, X., Cai, H.,2013. Industrial sector-based volatile organic compound (VOC) source profilesmeasured in manufacturing facilities in the Pearl River Delta, China. Sci. TotalEnviron. 456e457, 127e136.